Research map / Label-efficient and robust learning / Self-supervised and contrastive
TopoSRL: Topology preserving self-supervised Simplicial Representation Learning
Hiren Madhu, Sundeep Prabhakar Chepuri
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use proposed augmentation (add open, remove closed simplices) instead of random augmentation (ablation) | TopoSRL · contact-high-school order 3 | accuracy: 0.62 -> 0.92 | Improved |
| use proposed augmentation instead of random (ablation) | TopoSRL · contact-primary-school order 3 | accuracy: 0.36 -> 0.61 | Improved |
| use proposed augmentation instead of random (ablation) | TopoSRL · senate-bills order 3 | accuracy: 0.52 -> 0.72 | Improved |
| also remove closed simplices instead of only adding open simplices (ablation) | TopoSRL · contact-high-school order 3 | accuracy: 0.81 -> 0.92 | Improved |
| also remove closed simplices instead of only adding open simplices (ablation) | TopoSRL · contact-primary-school order 3 | accuracy: 0.51 -> 0.61 | Improved |
| only add open simplices instead of random augmentation (ablation) | TopoSRL · contact-high-school order 3 | accuracy: 0.62 -> 0.81 | Improved |
| include simplices of order 4 instead of order 3 only (ablation) | TopoSRL · contact-high-school | accuracy: 0.92 -> 0.85 | Hurt |
| include simplices of order 4 instead of order 3 only (ablation) | TopoSRL · contact-primary-school | accuracy: 0.61 -> 0.52 | Hurt |
| use TopoSRL (SSL) instead of supervised MPSN with 20% train labels | MPSN · contact-high-school | accuracy: 0.74 -> 0.79 | Improved |
| use TopoSRL instead of supervised MPSN with 40% train labels | MPSN · contact-high-school | accuracy: 0.80 -> 0.84 | Improved |
| increase training split from 20% to 80% | TopoSRL · contact-high-school | accuracy: 0.79 -> 0.92 | Improved |
| use TopoSRL instead of MPSN | MPSN · contact-primary-school node classification | accuracy: 0.79 ± 0.06 -> 0.61 ± 0.05 | Hurt |
Closest papers by meaning
- TopoGCL: Topological Graph Contrastive Learning (AAAI 2024 · similarity 0.87)
- TopER: Topological Embeddings in Graph Representation Learning (NeurIPS 2025 · similarity 0.86)
- Topological Neural Networks go Persistent, Equivariant, and Continuous (ICML 2024 · similarity 0.85)
- Cover learning for large-scale topology representation (ICML 2025 · similarity 0.85)
- Topological Blindspots: Understanding and Extending Topological Deep Learning Through the Lens of Expressivity (ICLR 2025 · similarity 0.85)
- TopInG: Topologically Interpretable Graph Learning via Persistent Rationale Filtration (ICML 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
